OpenAI doesn’t train GPT-5, reassures CEO

OpenAI CEO Sam Altman confirmed that the company is currently not providing GPT-5 training and will not be training for a while. The announcement followed rumors that a successor to the GPT-4, which was released in March, is in development.

With the increasing development of artificial intelligence, there are more and more concerns about the security and control of these systems. The open letter, signed by technology company executives, called for a pause in the development of artificial intelligence systems because of the possibility that such systems could be dangerous and difficult to control.

Altman discussed the security concerns surrounding artificial intelligence (AI) at an event at MIT. OpenAI continues to develop the latest version of ChatGPT and “does things outside of GPT-4 with all sorts of security issues that are important to address and are completely excluded from the charter.”

AI complexity

The CEO stated that the discussion is much more complex than stopping the development of more advanced systems. He explained that this is not necessarily true for AI systems where progress is much more complex and cannot be measured by version numbers alone.

Often the release sequence gives the impression of linear improvements. However, this may actually just be a marketing strategy by companies with no practical impact on the technology’s capability.